X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/feb5020b37d7d3ba4005a8bac6f4efece4ce4b8c..725f8b2d4cc43b8401946dfed69d22cfe40810fc:/youtube_dl/extractor/spiegel.py diff --git a/youtube_dl/extractor/spiegel.py b/youtube_dl/extractor/spiegel.py index 13c8640..fc995e8 100644 --- a/youtube_dl/extractor/spiegel.py +++ b/youtube_dl/extractor/spiegel.py @@ -1,45 +1,187 @@ +# coding: utf-8 +from __future__ import unicode_literals + import re -import xml.etree.ElementTree from .common import InfoExtractor +from .nexx import ( + NexxIE, + NexxEmbedIE, +) +from .spiegeltv import SpiegeltvIE +from ..compat import compat_urlparse +from ..utils import ( + extract_attributes, + unified_strdate, + get_element_by_attribute, +) class SpiegelIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?spiegel\.de/video/[^/]*-(?P[0-9]+)(?:\.html)?(?:#.*)?$' - _TEST = { - u'url': u'http://www.spiegel.de/video/vulkan-tungurahua-in-ecuador-ist-wieder-aktiv-video-1259285.html', - u'file': u'1259285.mp4', - u'md5': u'2c2754212136f35fb4b19767d242f66e', - u'info_dict': { - u"title": u"Vulkanausbruch in Ecuador: Der \"Feuerschlund\" ist wieder aktiv" + _VALID_URL = r'https?://(?:www\.)?spiegel\.de/video/[^/]*-(?P[0-9]+)(?:-embed|-iframe)?(?:\.html)?(?:#.*)?$' + _TESTS = [{ + 'url': 'http://www.spiegel.de/video/vulkan-tungurahua-in-ecuador-ist-wieder-aktiv-video-1259285.html', + 'md5': '2c2754212136f35fb4b19767d242f66e', + 'info_dict': { + 'id': '1259285', + 'ext': 'mp4', + 'title': 'Vulkanausbruch in Ecuador: Der "Feuerschlund" ist wieder aktiv', + 'description': 'md5:8029d8310232196eb235d27575a8b9f4', + 'duration': 49, + 'upload_date': '20130311', + }, + }, { + 'url': 'http://www.spiegel.de/video/schach-wm-videoanalyse-des-fuenften-spiels-video-1309159.html', + 'md5': 'f2cdf638d7aa47654e251e1aee360af1', + 'info_dict': { + 'id': '1309159', + 'ext': 'mp4', + 'title': 'Schach-WM in der Videoanalyse: Carlsen nutzt die Fehlgriffe des Titelverteidigers', + 'description': 'md5:c2322b65e58f385a820c10fa03b2d088', + 'duration': 983, + 'upload_date': '20131115', + }, + }, { + 'url': 'http://www.spiegel.de/video/astronaut-alexander-gerst-von-der-iss-station-beantwortet-fragen-video-1519126-embed.html', + 'md5': 'd8eeca6bfc8f1cd6f490eb1f44695d51', + 'info_dict': { + 'id': '1519126', + 'ext': 'mp4', + 'description': 'SPIEGEL ONLINE-Nutzer durften den deutschen Astronauten Alexander Gerst über sein Leben auf der ISS-Station befragen. Hier kommen seine Antworten auf die besten sechs Fragen.', + 'title': 'Fragen an Astronaut Alexander Gerst: "Bekommen Sie die Tageszeiten mit?"', + 'upload_date': '20140904', } - } + }, { + 'url': 'http://www.spiegel.de/video/astronaut-alexander-gerst-von-der-iss-station-beantwortet-fragen-video-1519126-iframe.html', + 'only_matching': True, + }, { + # nexx video + 'url': 'http://www.spiegel.de/video/spiegel-tv-magazin-ueber-guellekrise-in-schleswig-holstein-video-99012776.html', + 'only_matching': True, + }] def _real_extract(self, url): - m = re.match(self._VALID_URL, url) - video_id = m.group('videoID') + video_id = self._match_id(url) + webpage, handle = self._download_webpage_handle(url, video_id) - webpage = self._download_webpage(url, video_id) + # 302 to spiegel.tv, like http://www.spiegel.de/video/der-film-zum-wochenende-die-wahrheit-ueber-maenner-video-99003272.html + if SpiegeltvIE.suitable(handle.geturl()): + return self.url_result(handle.geturl(), 'Spiegeltv') + + nexx_id = self._search_regex( + r'nexxOmniaId\s*:\s*(\d+)', webpage, 'nexx id', default=None) + if nexx_id: + domain_id = NexxIE._extract_domain_id(webpage) or '748' + return self.url_result( + 'nexx:%s:%s' % (domain_id, nexx_id), ie=NexxIE.ie_key(), + video_id=nexx_id) + + video_data = extract_attributes(self._search_regex(r'(]+id="spVideoElements"[^>]+>)', webpage, 'video element', default='')) - video_title = self._html_search_regex(r'
(.*?)
', - webpage, u'title') + title = video_data.get('data-video-title') or get_element_by_attribute('class', 'module-title', webpage) + description = video_data.get('data-video-teaser') or self._html_search_meta('description', webpage, 'description') - xml_url = u'http://video2.spiegel.de/flash/' + video_id + u'.xml' - xml_code = self._download_webpage(xml_url, video_id, - note=u'Downloading XML', errnote=u'Failed to download XML') + base_url = self._search_regex( + [r'server\s*:\s*(["\'])(?P.+?)\1', r'var\s+server\s*=\s*"(?P[^"]+)\"'], + webpage, 'server URL', group='url') - idoc = xml.etree.ElementTree.fromstring(xml_code) - last_type = idoc[-1] - filename = last_type.findall('./filename')[0].text - duration = float(last_type.findall('./duration')[0].text) + xml_url = base_url + video_id + '.xml' + idoc = self._download_xml(xml_url, video_id) - video_url = 'http://video2.spiegel.de/flash/' + filename - video_ext = filename.rpartition('.')[2] - info = { + formats = [] + for n in list(idoc): + if n.tag.startswith('type') and n.tag != 'type6': + format_id = n.tag.rpartition('type')[2] + video_url = base_url + n.find('./filename').text + formats.append({ + 'format_id': format_id, + 'url': video_url, + 'width': int(n.find('./width').text), + 'height': int(n.find('./height').text), + 'abr': int(n.find('./audiobitrate').text), + 'vbr': int(n.find('./videobitrate').text), + 'vcodec': n.find('./codec').text, + 'acodec': 'MP4A', + }) + duration = float(idoc[0].findall('./duration')[0].text) + + self._check_formats(formats, video_id) + self._sort_formats(formats) + + return { 'id': video_id, - 'url': video_url, - 'ext': video_ext, - 'title': video_title, + 'title': title, + 'description': description.strip() if description else None, 'duration': duration, + 'upload_date': unified_strdate(video_data.get('data-video-date')), + 'formats': formats, } - return [info] + + +class SpiegelArticleIE(InfoExtractor): + _VALID_URL = r'https?://(?:www\.)?spiegel\.de/(?!video/)[^?#]*?-(?P[0-9]+)\.html' + IE_NAME = 'Spiegel:Article' + IE_DESC = 'Articles on spiegel.de' + _TESTS = [{ + 'url': 'http://www.spiegel.de/sport/sonst/badminton-wm-die-randsportart-soll-populaerer-werden-a-987092.html', + 'info_dict': { + 'id': '1516455', + 'ext': 'mp4', + 'title': 'Faszination Badminton: Nennt es bloß nicht Federball', + 'description': 're:^Patrick Kämnitz gehört.{100,}', + 'upload_date': '20140825', + }, + }, { + 'url': 'http://www.spiegel.de/wissenschaft/weltall/astronaut-alexander-gerst-antwortet-spiegel-online-lesern-a-989876.html', + 'info_dict': { + + }, + 'playlist_count': 6, + }, { + # Nexx iFrame embed + 'url': 'http://www.spiegel.de/sptv/spiegeltv/spiegel-tv-ueber-schnellste-katapult-achterbahn-der-welt-taron-a-1137884.html', + 'info_dict': { + 'id': '161464', + 'ext': 'mp4', + 'title': 'Nervenkitzel Achterbahn', + 'alt_title': 'Karussellbauer in Deutschland', + 'description': 'md5:ffe7b1cc59a01f585e0569949aef73cc', + 'release_year': 2005, + 'creator': 'SPIEGEL TV', + 'thumbnail': r're:^https?://.*\.jpg$', + 'duration': 2761, + 'timestamp': 1394021479, + 'upload_date': '20140305', + }, + 'params': { + 'format': 'bestvideo', + 'skip_download': True, + }, + }] + + def _real_extract(self, url): + video_id = self._match_id(url) + webpage = self._download_webpage(url, video_id) + + # Single video on top of the page + video_link = self._search_regex( + r'\s*.*?url\s*=\s*"([^"]+)"', + webpage) + entries = [ + self.url_result(compat_urlparse.urljoin( + self.http_scheme() + '//spiegel.de/', embed_path)) + for embed_path in embeds] + if embeds: + return self.playlist_result(entries) + + return self.playlist_from_matches( + NexxEmbedIE._extract_urls(webpage), ie=NexxEmbedIE.ie_key())